►
From YouTube: Kubernetes Community Meeting 20170518
Description
We have PUBLIC and RECORDED weekly video meetings every Thursday at 10am US Pacific Time.
https://docs.google.com/document/d/1VQDIAB0OqiSjIHI8AWMvSdceWhnz56jNpZrLs6o7NJY
Local development demo; Releases; SIG-API-Machinery, SIG-AWS, SIG-Openstack; SIG-Azure Proposal; CNCF update
A
A
B
Okay,
so
my
name
is
Yvonne.
Should
you
know
if
I
work
for
marriages
and
I'm
going
to
give
a
demo
projector
I've
been
working
for
some
time?
It's
the
current
docker
bias-based
local
cluster
solution,
and
it
was
discussed
a
quite
a
bit
on
a
single
cluster
life
circle
and
I
will
give
a
demo
on
yesterday's
country
box
meeting.
B
So
basically
what
this
project
is
about.
We
all
know
that
there
are
a
couple
of
widely
used
tools
for
local
development,
namely
a
mini
cube
and
local
coaster,
but
with
mini
Cuba
mini
Cuba's,
currently
not
very
well
suited
for
development
of
kinetics
itself,
because
it's
rather
hard
to
update
the
binaries
from
local
kubernetes
source
and
both
both
of
these
tools
are
also
run
single
mode
clusters,
and
sometimes
you
may
want
to
have
a
real
multi
node
cluster
and
you
may
want
to
run
enter
in
this,
for
it.
B
B
So
here
I
started
it
in
bills
mode.
It
will
be
of
Cuba
natesan,
Cuba,
GM
excusing
ppm
to
deploy
multi-node
posture
and
ochre
and
ochre
and
Here
I
am
starting.
It
is
pre
built
image.
So
what
are
scenarios
that
this
pool
covers?
For
example,
you
are
working
on
Cuban
lettuce
and
you
may
want
to
run
cement
in
tests
or
write
some
that
require
multiple
notes,
and
maybe
your
machine
is
not
very
powerful
and
you
don't
want
you
to
use
paid
clouds.
B
B
B
If
you
are
using
math,
you
likely
already
ran
in
docker
and
if
you
say
use
mini
cube,
you
will
have
to
run
another
VM
which
then
cause
its
own
set
of
problems.
So
now
we
see
that
this
is
kind
of
a
faster
machine
and
here
kubernetes
was
built
and
the
quarter
was
started
with
freshly
built
binaries
and
now
I
can
try
to
run
an
end-to-end
test
against
it.
Sorry
Rumple
so
here
are
also
helper.
B
Scripted
soaps
want
to
run
end-to-end
tests,
it
builds
into
a
test
runner
and
it's
quite
handy
when
development
and
ten
tests
and
running
them
locally.
While
completes,
we
see
that
here
the
Questor
already
started.
Actually
the
first
cluster
startup
takes
quite
some
time,
but
if
it
makes
a
snapshot
of
Wester
configuration,
so
if
I
start
it
again,
it
will
take
maybe
under
50
seconds
here
and
here.
If
I
have
been
to
restart
it,
it
will
take
maybe
40
seconds
or
less
so.
B
B
Standard
tools
may
cause
some
problems.
Also,
it
may
be
handy
if
you
say,
wants
to
change
something
in
queue
by
DM
and
test
it
quickly.
So
it
may
be
also
handy
here
now
concerning
this
CI.
What
I
was
thinking
about
actually
keep
IDM
VND
cluster
as
it
sounds
CI.
One
of
its
features
is
that
cluster,
despite
being
the
current
local
after
its
kind
of
realistic
in
existence,
it,
for
example,
would
make
me
run
multiple
CNI
implementations
such
as
flannel
calico
and
we've
see
41.6
out
of
some
flakes
flakes
with
one
Evan
people.
B
They
do
work
but,
like
often
on
say,
I
saw
a
temporal
disable
them,
but
I
think
I'll
fix
it
shortly.
Here
is
the
multiple
test
cases
of
genda
cluster
running
on
Travis,
it's
some
of
them
use,
previewed
binaries
of
kubernetes
and
some
of
them
beauty
benefits
from
source.
You
can
tell
from
few
times
here's
another
example
of
the
nd
quarter
usage.
It's
work
that
projects
it's
Sarah,
implementation,
they're,
trans
VMs.
There
was
a
demo
courtesy
here
some
time
ago.
B
Actually-
and
here
you
can
see
that
you
can
download
the
simple
script
and
run
it
to
play
around
this
work
at
the
same
studio
in
discrete
and
also
you
gain
de
cluster
is
used
for
CIF
working
project.
The
main
it
is
the
project
says
it's
rough
edges
currently
and
probably-
downside
is
that
it's
mostly
collection
of
shell
sticks
but
I'm
going
to
rewrite
it
and
go
and
I
made
a
proposal.
B
I
put
it
in
to
the
community
repository
about
the
new
tool
based
on
the
same
idea,
but
basically
written
like
mango
and
with
more
clean
configuration
structure,
so
basically
see
here
and
then
test
passed.
So.
B
Basically,
the
so,
except
maybe
I
can
just
restart
it
without
a
build
and
see
if
it
indeed
will
happen
rather
quickly,
it
may
be
made
even
more
fast,
because
cue
EDM
is
not
you're
done
when
you
restart
the
cluster,
just
natural
teeth,
reused
and
it
spends
actually
most
been
waiting
for
the
cluster
over
when
the
cluster
is
ready.
So
you
don't
have
some
non
ready
pods
when
it
completes.
B
B
C
A
B
C
C
C
D
C
D
So
so
we
only
have
those
people
we
found.
Those
treatment
in
civil
bills
for
the
most
is
called
the
easier
types
itself
or
it
is
the
path
infrastructure
business.
We
are
in
the
middle
of
the
some
of
the
pencil
fields
changing
to
the
new
infrastructure
problems
for
should
be
back
on,
and
but
we
easily,
we
do
identify
a
really
nice
soccer,
which
you
want,
the
cube
epoxies
we
start
and
pretty
much
just
never
get
or
come
functionally.
D
So
thanks
to
sig,
sig
network
hard
work
and
the
vultures
are
offended
coming
and
now
anything
looks
clean
so
far
and
if
you
have
the
thickness,
but
it
is
not
proper
for
the
our
body
needs.
So
there
is
the
way
click
on
the
build
this
morning,
so
we'll
send
the
answer
updates.
It
should
make
us
the
nation's
hope.
Communica
has
to
lot
of
it
new
builders.
Another
thing
is
that
we
are
working
on
the
up
up
with
pencil.
Great
we
still,
we
put
those
some
other
operates,
has
to
become
our
school
types.
D
D
So
we
should
have
the
106
upwards
to
the
master
branch
of
shortening,
and
then
we
plan
to
choose
one
I'm
in
syntagmatic
we
plan
to
choose
one
and
as
blocker
the
preset
news
broker
so
come
to
the
contrary,
targeted
for
the
only
upgrade
master
and
the
founder
one
who
sticks
to
the
master
branch.
So
that's
one
for
the
first
appendix
so
where
we
maybe
have
more
to
excess
offer
so
and
I
end
up
what
elements
tonight.
D
So
lenient
team
is
talking
about
the
new
process,
about
the
issues
and
the
feature,
issues
or
recognize
each
artifact,
and
there
are
certain
Picasso
and
it
doesn't
towards
field,
and
so
next
week
I
think
the
diode
is
going
to
give
the
demo
for
that
one,
and
so
so
so
lots
of
things
master
things.
It
is
called
a
phase
of
schedule,
de
I'm,
June.
First
so
far
does
not
change
and
hours.
I
hope,
they'll
communicate.
E
Just
one
not
from
me,
you're
getting
two
features:
it's
yet
another
friendly
reminder
that
if
you
had
a
future
owner,
please
take
your
picture
description.
I've
been
grooming
throughout
that
globe
and
I'm
that
most
of
the
year
updated
them
all
about.
Some
of
the
teachers
are
still
not
updated
and
through
the
difficult
to
find
privileged
edges
of
the
features.
So
please,
if
your
own,
in
the
future
407
there's
a
day.
That's
all
for
me.
Yeah.
F
I
just
want
to
I
want
to
double
down
on
what
cor.
Just
said
from
the
fanfic
perspective,
we
use
the
features
repo
to
inform
things
without
each
other's
work.
We
also
use
it
for
the
release
notes
and
we
use
it
for
the
blog
post.
So
if
you're
expecting
to
be
part
of
any
of
those,
please
provide
a
description
that
can
be
used
in
release,
notes
also
for
the
PM
sig
we're
going
to
be
inviting
feature
owners
from
the
different
things
to
come
and
give
free
demos,
and
so
we'll
have
a
schedule
for
that.
G
Anthony,
so
for
anyone
who
didn't
see
the
announcement
in
163,
we
introduced
a
regression
that
could
cause
some
pots
to
not
start
that
used
to
work
before
so
we
sent
out
an
announcement
saying,
please
don't
use
163
yet
and
we're
working
on
a
fix
that
we're
going
to
put
out
a
quick
164
at
the
end
I'm
taking
a
little
bit
longer
than
we
thought
to
get
that
fix
submitted.
But
that
is
now
cherry-picked
into
the
one.
Six
branch
and
what's
next,
is
to
cut
164.
G
But
we
plan
to
hide
one
to
explore
only
with
the
fixes
for
that
critical
bug
and
one
other
regression
that
came
up
with
a
cubelet
panic.
That
also
was
new
in
163.
So
those
two
bug
fixes
are
the
only
things
that
should
go
into
one
six:
four
and
that's
not
a
very
well
exercised
process.
So
that's
still
to
be
determined
how
long
that's
going
to
take
but
I'm
working
on
it.
Now.
C
A
How
many
one
okay?
Well
in
that
case,
let's
move
on
to
the
cig
reports,
so
Dan,
you
leave
a
report
on
machinery
and
there's
already
some
questions
in
the
notes,
so
we'll
take
those
as
soon
as
you
give
you
a
report,
you
can
take
the
questions.
First,
if
you
want
okay,
chao-zu
had
a
question
regarding
an
API
issue.
A
H
So
a
cig
API
machinery
is
focused
on
extension
points
for
this
release,
so
we're
working
hard
to
get
user
user
API
servers
aggregated
with
the
main
API
server,
we're
also
working
to
get
initializers
moved
out
of
process.
Initializers
are
one
of
the
mechanisms
that
we're
going
to
use
to
to
implement
admission
controllers
out
of
other
processes.
So
there's
a
bunch
of
stuff
that
you
can't
do
without
building
code
into
cube
api
server
and
the
goal
is
to
have
in
solid
alpha
or
beta
these.
These
extension
points,
so
you
can
build
stuff
out.
H
So
let
me
just
sorry:
I
don't
have
the
slides
or
anything,
but
I
would
like
to
pull
out
a
couple
things
that
are
from
our
notes
a
few
meetings
ago,
still
demoed
a
API
server,
SDK
proof
of
concept.
So
this
would
let
people
build
their
own
API
server
or
make
it
easier
to
build
your
own
API
server.
You
should
have
to
change
just
a
few
lines
of
code
instead
of
lots
of
lines
of
code.
H
Okay,
we've
had
a
lot
of
discussions
about
the
design
of
the
the
aggregation
layer
on
particular
what
sorts
of
network
traffic
it
may
send
and
what
are
the
requirements
for
network
traffic
from
control
plane
elements?
This
is
this
is
a
basically
control
plane.
Elements
need
to
function
in
multiple
different
Network
environments.
Sometimes
the
master
nodes
are
not
in
the
same
network
as
the
as
the
worker
nodes,
so
the
stuff
that
you
find
when
you're
starting
to
put
it
put
it
all
together,
see
the
other
thing
yeah.
H
So
we
have
solidified
the
proposals
for
the
admission
controllers
and
initializers
that
I
mentioned
earlier.
The
Clayton's
main
proposals,
demerged
Chawla,
has
a
design
which
is
probably
going
to
get
merged
today,
and
hopefully
we
have
that
operating
by
the
end
of
the
future
freeze
or
before
the
feature
freeze
rather-
and
we
also
had
a
lively
round
of
discussion
last
time
about
accepting
more
clients,
so
I
think
you
can
look
for
more
clients
in
other
languages
in
the
future.
H
Sorry,
I'm
distracted
by
the
the
chat,
I'll
answer
that
when
I
get
to
the
end
of
what
I
was
going
to
say,
we
have
a
long-standing
desire
in
API
machinery
to
package
up
our
code
in
ways
that
respect
the
different
users
of
it.
So
if
somebody
wants
a
client,
they
should
be
able
to
get
a
client
without
venn
during
the
entire,
like
you
shouldn't
need
to
vendor
cubelet
to
get
a
client.
If
you
want
to
write
an
API
server,
you
should
need
to
vendor
cubelet
to
get
a
a
PA
server.
H
So,
unfortunately,
our
code
has
a
bunch
of
little
spaghetti
strands
that
that
cross
components
and
the
final
thing
that's
blocking
us
from
completing
this
process
is
the
API
types
themselves.
So
right
now,
the
client
code
library
has
a
copy
of
the
API
types,
which
is
totally
fine
if
you're,
using
only
the
client
library
or
only
code
in
the
main
repository.
But
people
need
to
compose
our
various
components
and
if
you
try
to
import
both
API
server
and
client,
suddenly
you
discover
that
you
have
two
copies
of
the
API
types
and
they
don't
work
together.
H
So
the
final
step
in
this
process
is
to
actually
split
the
API
types
themselves
into
a
repo
that
both
the
client,
libraries
and
the
other
libraries
can
depend
on,
and
at
that
point
we
can.
We
can
clear
up
a
lot
of
the
weird
stuff
in
the
staging
directory.
I'm
sure
you've
all
noticed
that
some
of
the
files
in
the
same
directory
are
canonical
in
the
staging
directory,
and
some
of
them
are
like
copied
and
assembled
from
other
places,
and
that
is
super
confusing,
so
to
make
a
long
story
short.
H
The
current
plan
of
record
is
that,
after
the
after
the
the
the
feature,
the
code
freezes
over
Chow
has
done
a
bunch
of
work
to
do
that
separation
and
we
think
that's
the
least
disruptive
time
to
actually
go
ahead
and
put
that
in
the
in
the
repository.
So
if
you
have,
if
you
have
thoughts
on
that,
the
issue
posted
in
is
probably
the
best
place
to
discuss
that
it
yeah.
So
hopefully
that
makes
sense
we.
So
we
plan
to
do
this
as
like
the
first
thing
after
the
freeze
is
lifted.
H
Any
thoughts
on
the
build,
your
own
API
surfers
tutorial,
so
I
have
to
say:
I'm
focused
right
now
on
making
it
possible
at
all
so
I
think
we're
going
to
need
tutorials
and
SDKs
and
pretty
documentation
before
we
can
exit
beta,
but
I.
Think
at
the
moment
we
need
to
focus
on
making
making
this
possible.
That's
that's
the
part
that
has
to
work
before
the
feature
freeze
after
the
feature
freeze
that
we
can
think
about
making
it
fixing
things
that
aren't
blocked
on
feature
freezes
like
tutorials
and
fills
a
fills.
Still
thing.
H
Okay
is
that
did
I
cover
all
the
questions,
there's
also
a
question
about
API
API,
aggregation
and
1/7.
Yes,
so
the
current
plan
is
that
API
aggregation
will
be
available
in
1.7.
The
aggregation
layer
will
be
built
into
cube,
API
server
yeah,
so
the
there's
a
unfortunately
I,
attracted
in
a
org
level
project
which
apparently
apparently
non-work
members
don't
have
access
to.
But
if
you
are
organ
ever
you
can
look
at
the
aggregation,
github
project
to
see
what
are
the
remaining
items
and
what
we're
working
on.
H
Let's
see
someone
asks.
Is
this
motivated
by
kubernetes
core
strategy,
I
think
the
answer
to
that
is
got
to
be.
Yes,
we
we
can't.
We
can't
scale
the
the
current
process
where
everybody
who
wants
an
API
build
something
into
a
cube,
API
server.
Just
doesn't
it
isn't
scalable,
so
we've
got
it.
We've
gotta
separate
this
we've
got
to
unblock
people
who
want
to
on
experiment.
We've
got
to
make
it
able
make
it
possible
to
turn
api's
on
and
off
without
invasive
code
changes.
So,
yes,.
A
J
And
apologies
they're
doing
drilling
in
my
building
so
for
any
reason
like
it
sounds
like
there's
this
huge
shaking,
that's
because
of
the
drilling
but
yeah
so
updates
from
say,
Gabe,
AWS,
first
sort
of
a
meta
update
is
that
we
have
been
for
the
for
the
past
couple
of
months
doing
this
very
specific
format
of
sig
meetings,
where
we
always
thought
worth
intros
and
updates
from
everyone
in
the
in
the
in
the
group
on
progress
or
or
any
of
the
challenges
based
and
they
overcame.
J
In
the
past
couple
of
weeks
we
assign
roles
in
beginning
to
like
note-taking
stuff,
and
then
we
always
have
demos,
of
course,
and
we're
thinking
about
introducing
presentations
from
different
members
on
how
they're
implementing
troop
entities
on
AWS
there
are
companies
kind
of
give
real-world
use
cases,
and
this
has
been
pretty
well
received.
So
that
was
just
a
student
update
on
on
our
format
and
particularly
to
surround
the
intros
every
time.
I
think
it.
J
J
The
first
is
from
hunting
at
the
londo
on
his
work,
around
external
DNS
incubator
project,
and
the
second
is
from
Josh
from
core
OS
who's,
giving
a
demo
of
the
alb
work
on
AWS
work
that
we've
been
doing
with
Ticketmaster.
So
this
is
the
updates
from
us
and
Brandon
just
link
to
add
to
the
queue
I
am.
J
A
E
E
I
was
into
sick
and
what
conversation
specifically
over
here,
but
I
don't
want
sex,
and
so,
first
of
all,
I'd
like
to
say
thank
you
for
all
the
people
who
current
OpenStack
summit
on
behalf
of
kubernetes
community
I
was
really
surprised
to
see
so
many
faces
from
keeping
at
his
community,
and
especially
people
who
joined
our
cabinet
is
dated
happened
on
the
second
day
of
OpenStack
summit.
Also,
I
would
say
that
compared
to
the
previous
summits,
the
more
birthdays
and
a
world
proportion
of
communities
related
talks
to
component
as
non
related.
E
Folks,
it's
much
higher,
so
I
would
say
that
I
can
fit
it
like
almost
a
cup,
but
approximately
that
amount
we
head
of
cabinet
is
given
address
related
talks
at
OpenStack,
some
at
the
general
track,
not
speaking
about
kubernetes
day.
That
has
been
fully
dedicated
to
the
Panetta's.
So
so
we
also
have
several
several
mediums
at
open
sticks
summit
and
one
of
them
were
skipping.
A
disciplines
take
on
also
an
open,
steady
fixation
during
forms
of
the
stiphu
can
bring
us
no
feedback
from
you
say:
oh.
K
Yes,
so
that
there's
a
lot
of
notes
need
the
pad
from
that
session,
the
main
thing
that
came
up
and
is
kind
of
a
theme
I
think
of
some
of
the
keynote
demos
during
the
week
as
well.
There
is
a
desire
among
the
operators
to
have
a
West
monolithic
approach
to
the
cloud
provider
integration
we
have
currently
so
be
able
to
pick
and
choose
which
of
the
OpenStack
services
they
use
with
kubernetes.
K
So
on
that
note,
the
main
kind
of
feature
work
is
still
was
still
trying
to
achieve
within
the
OpenStack
stick
hours
around
the
externalization
of
the
cloud
provider.
Absolute
ongoing
efforts.
Teams
in
particular
has
at
least
bit
of
mirror
of
the
code
across
solid
OpenStack
infrastructure
and
here
in
research,
we're
working
on
how
to
set
up
test
infrastructure
around
that,
but
we
still
have
a
fair
bit
of
work
to
do
there
and
they
are
looking
for
additional
contributors
which
we're
trying
to
find
them.
K
We
are
also
trying
to
find
some
additional
owners
for
the
cinder
integration,
because
currently
I
think
this
is
the
one
person
from
six
storage
on
there.
Absolutely
I
have
another
stack
focus
person
on
that
as
well
cinder
being
to
the
earlier
point
of
operating
on
the
monolith.
Cinder
is
fully
the
service
to
it
closest
to
being
able
to
use
individually.
There
is
a
separate
volume
implementation
for
that
in
a
yes
conversion,
I
think
will
request,
but
pretty
close
to
English
for
memory.
Wow
I
think
those
it's
the
second
day.
K
He
was
comment
that
there
was
a
lot
of
Q&A
so
later
discussion,
a
lot
of
momentum
around
the
communities
to
assist
one
in
romanette
services,
aseptically
via
tech,
home
project
and
cookies,
and
then
growing
momentum
around
this
less
monolithic
approach.
So
you
better
lock
them
as
showcasing
using
just
one
individual
purpose
next
service
with
kubernetes
or
a
subset
of
what
we
would
normally
consider
the
course
excellence
of
student.
A
L
But
since
the
time
that
happened
and
now
there's
been
a
tremendous
amount
of
motion
forward
in
terms
of
the
azure
container
service
and
just
generally
crew,
Nettie's
love
across
the
greater
whole
of
Microsoft,
which
is
super
cool
because
it
represents
I,
think
a
huge
untapped
pool
of
resources
that
we'll
be
able
to
hopefully
channel
into
the
project.
So
I
feel
like
this.
L
Cig
is
an
important
bridge
for
those
activities
and
I
think
it's
time
to
look
into
this
and
one
caveat
because
I
know:
there's
concern
about
overlap
with
the
potential
of
other
medicines
related
to
cloud
providers
in
general
I
put
in
the
proposal
because
I'm
not
a
fan
of
six
brawl.
Let's,
if
there's
something
better
that
comes
along
that
meets
everybody's
needs
sake,
AWS,
like
GK
or
whatever
those
look
like.
L
We
can
definitely
look
at
a
meta
sig
for
all
those
and
and
deprecated
our
sig.
If
that
happened,
because
I
don't
want
to
keep
sitting
around
just
because
they're
precious
good.
So
what
I'd
be
looking
for
is
anybody
who
has
any
strong
objections
or
feels
like
it's?
Not
the
right
move,
and
we
can
talk
about
that,
but
otherwise
I
would
love
to
get
a
lazy
consensus
approval
for
it.
So
we
can
start
for
it
on
check
it
out
all
the
organised
I.
Don't.
I
Know
the
details
of
this,
but
wasn't
there
a
movement
to
have
a
big
cloud
where
today,
I
guess,
my
question
is:
if
we
had
like
a
spigot
of
us,
a
ciggie
ze:a,
a
bezoar,
a
cig,
you
know
whatever
the
new
cloud
provider
are
and
they
silo
of.
Are
we
not
going
to
be
able
to
leverage
the
you
know
the
ability
for
them
to
you
know
to
solve
problems
together
and
are
we
going
to
creating
an
opportunity
or
a
situation
where
we're
duplicating
effort?
There
wasn't
a
cloud
proposal
or
what
happened
to
that
I'm
just
curious.
I
L
M
My
concern
is
that
by
creating
more
and
more
sick
than
you
already,
the
results
already
seek
AWS
and
there
will
be
less
incentives
to
converge
later
on,
because
it'd
be
easier
to
do
something
propriety
for
specific
cloud
and
I
mean
the
promise
of
kubernetes
is
that
you
do
have
the
same
experience
and
it's
true
that
and
it's
about
the
portability
of
workers,
but
with
the
cube
idea.
For
example,
we
try
to
do
an
infinitude
also
before
the
muscle-up
stirs
it's
about
like
doing
something
bigger
for
the
community
in
kubernetes
versus
each
cloud.
N
M
Understand,
but
you
know,
I
can
look,
for
example,
what
Daniel
presented
and
the
investment
it's
real
and
making
right
now,
for
example,
on
accessibility.
This
is
something
for
the
bright
future
of
kubernetes
and
something
that
we
didn't.
We
could
have
maybe
gotten
away
without
doing
that.
So
how
are
we
going
to
make
sure
that
we'll
get
to
this
well.
N
I
guess
this
is
a
very
different
thing.
Right,
like
extensibility,
is
that
our
total
new
matrix,
not
a
shrink
where
we
already
have
users,
we
already
have
existing
work
done.
We
already
have
people
who
need
support
and
a
place
for
them
to
discuss
as
well
as
several
other
parallel
groups.
I
I
think
that
at
a
project
wide
level,
if
we
decide
that
you
know,
we
need
to
have
some
other
Cygnet
at
something
that
someone's
going
to
need
to
lead
out,
but
I
think
that
pinning
on
creating
a
new
sig.
N
M
Is
a
matter
I
advise
the
accessibility,
because
it's
also
a
my
total
polarization
and
where
we
invest
engineers
and
everybody
has
users
and
I-
definitely
don't
want
to
hold
things
back
and
we
can
and
it
doesn't
seem
like
the
Leadership
Summit
and
if
it's
the
right
place
to
do
a
PSO.
No.
But
if
you
won't
take
any
action
about
alignment
of
those
six,
then
nothing
will
happen.
We
know
that
you
know
this
and.
N
N
N
L
O
D
I
just
want
I,
don't
have
the
opinion
about
this
particular
snake
and
but
I
do
have
the
observation.
So
we
have
to
create
a
pound
of
the
six
and
then
I'm
really
really
worried
about
the
coordination.
It
is
no
matter
if
I
stood
in
each
manager.
Oh,
it
is
just
regular
stick
next.
One
of
the
sick
leave,
for
example,
is
my
fault.
Next
is
the
windows?
Runtime
relations
have
approached
a
structure.
We
we
just
don't
have
the
bandwidth
that
I
have
to
handle.
D
So
we
suggest
the
big
windows,
non-chemical,
so
particular
profit
versus
not
really
alignment
and
the
we
just
recently
found
ok
well,
I
would
vision,
went
to
the
cig
windows
meeting
and
I
found
Oh.
Actually
they
are
rely
on
certain
things,
but
because
there's
no
coordination,
no
alignment
and
we
just
accidentally
to
remove
some
of
them
their
features.
And
then
we
have
to
do
some
extra
staff,
the
chipping,
but
in
mobile
Eugene
is
equal
to
ask.
Nobody
have
not
to
say.
Oh
here's
signal,
so
it
here's
the
Civic
figure
windows,
the
the
PR.
D
The
the
purpose
is
not
a
review
by
ASCII
is
not
review
by
ask
the
feature
claim
is
not
to
review
by
us,
but
the
code.
It
is
somewhere
and
it
is
the
only
way
it
allows
fabricated
code,
so
nobody
knows
and
a
battery
mode.
So
if
I
don't
go
to
the
signal
windows-
and
we
don't
know,
though
camping
you
guys,
nothing
happen,
yeah.
N
L
Can
I
actually
clarify
that
it's
not
I,
don't
see
any
contention
about
cig,
a
sure
its
contention
about
the
meta
layers
of
govern,
so
I
would
say
that
the
lazy
consensus
that
I've
heard
so
far
is
that
there
is
backing
behind
it
from
Korres.
How
do
people
Microsoft
and
then
what
not
that
it
seems
like
that
fight
is
fine
and
we
defer,
if
there's
a
sig
cloud,
we'll
be
the
first
ones
to
help
support
that
against
you.
So
could
people
who
support
sig
adarand
plan
to
participate?
L
P
P
If
you
have
thoughts
or
things
you'd
like
differently,
please
reach
out
to
me
and
I'm
happy
to
discuss
any
of
these
areas
with
you
or
new
things.
We
should
be
doing
or
things
we
should
you
differently.
So
CFCF
is
a
little
less
than
18
months
old.
We're
part
of
the
Linux
Foundation
kubernetes
is
the
original
project
that
we
host,
but
we're
now
up
to
nine
here
and
in
fact
the
container
network
interface
is
going
to
come
in
as
our
country
next
week,
and
these
are
platinum
members
that
support
us.
P
The
basic
idea
for
what
we're
doing
on
slide
3
is
that
we
try
and
support
these
projects.
We
think
a
lot
of
them
can
then
be
useful
in
commercial
products
that
can
provide
profits
to
companies
that
allows
them
to
reinvest
into
this
project
so
we're
trying
to
help
enable
this
ethos
system
by
creating
trust
between
different
participants.
In
the
case,
communities
CNCs
has
a
governing
board
which
has
made
new
vendors
and
they
have
a
marketing
committee.
But,
interestingly,
they
don't
have
the
ability
to
bring
in
new
projects.
P
That's
all
in
there's
a
nine
member
technical
Oversight
Committee
of
which
Brian
grants
on
a
call
is
a
member
and
several
other
well-known
folks,
they're
the
people
who
can
actually
approve
new
projects
and
look
for
what
kind
of
resources
those
and
provided
resource
to
those
projects.
We
have
an
end-user
board
and
some
other
parts.
Here's
the
list
on
slide
five
of
the
members
of
our
technical
oversight
committee.
Six
of
them
are
picked
by
the
governing
board,
one
by
the
end-user
board.
P
P
Pacific
time,
there's
also
an
open
mailing
list
that
I'd
encourage
you
to
participate
in
if
you're
not
too
flooded
by
traffic,
and
then
this
is
a
preview
of
some
work
that
we've
been
doing
at
CNCs,
to
talk
about
the
different
projects
and
particularly
kubernetes,
and
how
unusual
it
is,
and
so
this
is
a
log-log
chart
where
the
x-axis
shows
the
logarithmic
commits.
The
y-axis
is
the
number
of
requests
and
issues,
and
you
can
see
that
Linda's
Colonel,
the
top
right
is
in
some
category
but
separate
from
that
kubernetes.
P
Also
at
the
top,
is
really
far
away.
Indigent
is
a
log-log
chart.
So
a
little
bit
of
distance
has
a
huge
difference
between
these.
You
can
click
where
it
says:
interactive
version
to
then
be
able
to
hover
over
these
and
see
the
specific
statistics,
so
I
think
that
the
main
point
that
I
would
want
communicate
to
everybody
here
is
to
the
degree
that
you
feel
like
kubernetes
is
overwhelming
and
a
huge
amount
of
effort,
and
it's
incredibly
challenging
to
coordinate
between
1,600
developers
and
such
there's
very
good
reason
for
that.
P
I
would
also
quote
the
executive
director
of
the
Linux,
Foundation
or
parent
in
saying
that
Nettie's
is
the
Linux
of
the
cloud,
and
so
there's
just
a
ton
of
attention
and
excitement
over
this.
Yes,
in
a
lot
of
ways,
was
formed
in
order
to
help
that
community
in
different
ways,
so
I
did
want
to
address
a
real
specific
question,
though,
on
this
next
slide
number
seven
of.
Why
isn't
there
kubernetes
foundation
and
just
quickly
when
Google,
obviously
originally
kubernetes
in
2014
2015?
P
It
was
always
designed
to
be
an
open
source
project,
but
at
least
in
principle,
they
could
have
gone
for
different
paths
with
it,
at
least
from
more
from
most
close
to
most
open,
so
they
could
have
made
it
a
closed
source
project
like
Amazon
Elastic
container
services
that
you
had
to
become
a
customer
of
Google
Cloud
to
use
it
they
on
the
next
option.
More
open
could
have
had
an
open
source
project
like
golang,
which
is
obviously
a
very
successful
project.
All
of
us
use
it,
but
it's
still
under
Google's
control.
P
The
third
option
is
they
could
have
created
a
google
kubernetes
foundation.
That
would
mainly
have
just
been
home
of
kubernetes,
but
they
made
the
decision
and
then
they
brought
in
all
the
other
partners
from
around
the
community
and
agreed
to
create
this
cloud
native
computing
foundation
that
had
kubernetes
as
the
anchor
tenant.
But
the
whole
idea
was
to
support
a
constellation
of
projects,
and
the
belief
was
that
in
doing
so,
they
would
have
more
support
and
be
able
to
accomplish
more
so
why'd.
P
They
choose
that
scenario
because,
as
I
said
before,
we
believe
that
a
neutral
clone
builds
trusts,
increases
the
willingness
of
developers
to
collaborate
contribute
and
when
you
look
at
some
of
the
projects
that
we're
now
hosting
things
like
G
RPC,
that's
fun.
Welcome,
spend
constructor
container
network
interface,
that's
coming
in
next
week,
hopefully
container
storage
interface
in
the
next
couple
months,
container
D.
We
think
that
all
these
working
together
can
have
a
bigger
impact
and
spur
innovation.
P
So
moving
on
to
that
next
slide,
what
does
CF
do
for
kubernetes
I
did
want
to
just
walk
through
these
real
quickly,
so
the
most
important
thing
is
providing
a
neutral
hold
that
different
collaborators
can
work
together
and
know
that
there's
not
one
company
that
is
in
control
or
that
can
force
things.
What.
P
Yeah
but
you
can
see
my
screen
share
so
yes,.
P
Apologize
so
the
going
to
that
list
again
we
provide
services
like
license
scanning
of
all
the
court
and
the
vendor
code,
legal
guidance
on
top
patented
copyright
issues
and
I.
Think
one
of
the
things
that
is
in
common
with
a
lot
of
these
is
they're,
not
particularly
blatant
obvious
things
we're
trying
to
provide
a
set
infrastructure
services
and
kind
of
help
around
the
edges.
P
We're
not
normally
in
this
call
or
in
other
ways,
sort
of
jumping
up
and
down
screaming,
but
we
are
there
to
provide
support
when
it
when
is
requested
and
they're
offering
to
kinds
of
support.
So
another
area,
that's
going
to
become
more
prominent,
is
an
open
source,
curriculum,
a
training
program
and
then
a
certification
program.
That's
going
to
be
going
to
beta
test
next
month
in
June,
we
have
a
software
conformance
working
group.
That
is
insuring
that
when
people
say
kubernetes,
it
means
it
has
a
meaning
means
something
and
end
is
compatible.
P
We
provide
pressure
P,
our
analyst
relations
services.
We
run
to
con.
We
tend
external
conferences
and
and
raise
awareness
of
kubernetes
there,
and
then
we
are
now
taking
over
the
management
of
kubernetes,
io
I
will
say
essentially
other
than
slash
docs,
but
the
partner
page,
the
home
page,
the
kind
of
basic
intro
pieces
to
that
we're
hosting
and
funding
developer
activities
like
through
Benetti
support.
We
do
some
private
infrastructure
like
a
github
repo
for
the
community
security
team
and
then
we're
essentially
available
for
ad
hoc
activities.
P
So
this
there
was
a
request
from
a
group
of
folks
to
create
a
standard,
kubernetes
and
neutral
Nettie's
am
I
that
could
live
in
the
AWS
marketplace.
You
need
to
have
an
organization
to
back
that,
and
so
we
were
happy
to
do
that.
That's
underway,
then
moving
on
slide,
nine
sent
just
some
other
wins
where
we
finally
convinced
slack
to
give
us
free
access
to
the
full
version,
so
that
folks
would
do
search
and
get
access
to
the
archive.
P
We
would
say
that
we're
helping
to
try
and
reduce
some
vendor
tensions
and
we're
doing
things
like
taking
over
the
logistics,
but
not
the
content,
senator
burr
Nettie's
dev
summit.
We
also
provide
minority
scholarships
for
our
events,
so
there's
this
question
of.
Does
CFCF
market
kubernetes
effectively.
I
won't
go
through
all
these
details
and
just
say
that
it
is
something
that
we
work
very
hard
at
and
then,
when
you
go
through
slide,
11
I
think
you
can
see
some
of
the
consequences
here,
which
is
I.
P
Certainly
don't
in
any
way
want
to
have
to
take
credit
for
this
I
think
it's
through
Nate's
project
and
the
success
of
how
useful
it
is
that
deserves
it.
But
the
key
point
is
that,
as
kubernetes
has
come
into
CN
CF
and
then,
as
we've
been
running
coupon
and
doing
other
things
for
doing
the
presses
of
the
where
the
latest
releases
etc.
You
can
see
that
that
growth
has
continued
over
time
compared
to
some
comparable
projects
and
on
slide
12.
P
You
can
see
that
we're
now
edging
in
towards
OpenStack
in
terms
of
this
mod
share
on
13
is
some
slides
about
media
coverage,
showing
it
going
up
over
time.
14
is
a
share
voice,
as
you
can
see,
kubernetes
being
featured
more
and
more
saying,
the
15/16
is
actually
some
work
that
I
put
together.
Just
talking
about
how
popular
and
successful
kubernetes
is
and
different
ways
of
measuring
that,
and
then
17
is
looking
at
some
work
that
we're
doing
with
today's
meetup
groups
and
other
meetup
groups
and
I.
P
Think
there's
been
a
little
confusion
here
and
thinking
that
cloud
native
meetups
are,
since
you
have
meetups
of
different
than
kubernetes
groups.
All
that
happens
is
we
reach
out
to
the
existing
groups
out
there
encourage
them
to
register
as
a
SAN
staff
meeting
group
and
once
they
do,
that
we
provide
subsidized
food
beverage,
provide
swag
with
speakers,
bureau
and
other
kinds
of
services
and
then
slide.
18
is
probably
the
only
news
that
I'm
going
to.
P
P
We
will
have
to
come
on
Thursday
and
Friday
December,
7th
and
8th,
where
all
the
content
will
be
kubernetes
related,
so
either
raised
directly
or
how
these
other
projects
work
with
threads
slide.
19
is
just
a
look
at
the
punching
for
Berlin.
It
shows
that
we
were
already
a
huge
amount
of
it
was
turbine.
Eddie's
related,
so
like
20
is
showing
this
huge
growth
and
conference.
P
And
up
to
2000
and
then
sex
21
just
says
why
we
don't
do
some
things,
which
is
that,
although
we're
open
to
them,
we're
somewhat
hesitant
to
hire
product
managers
or
punch
managers
or
build
cops
or
other
folks,
because
we're
trying
not
to
compete
with
our
members
here.
So
we
are
working
with
the
two
Rene's
leadership
groups
and
we're
open
to
working
with
more
folks
to
try
and
help
provide
resources
and
focus
the
resources.
But
the
idea
for
the
foundation
has
never
been
to
be
that
the
foundation
itself
employs
three
contributors.
P
Last
two
slides
22
is
just
saying
that
we
do
have
meaningful
budget
available
today
and
we're
particularly
interested
and
focused
on
trying
to
help
some
of
the
infrastructure
issues
around
munchbot
and
prowl
other
things,
documentation
that
are
causing
challenges
for
the
kubernetes
community
that
and
we're
also
willing
to
take
over
budget
responsibility
for
confusing
integration,
confusion
or
registry
infrastructure.
The
challenges
that
we'd
be
looking
for
the
community
SIG's
to
develop
consensus,
create
a
transition
plan
and
that
hasn't
been
a
priority
for
folks.
Yet
slide.
23
is
really
my
final
thought,
which
is
that
I?
P
Really
we
at
st.
CF
at
the
Linux
Foundation
18,
the
governing
board,
cetera,
have
this
read
that
what
we're
doing
here
is
fundamentally
a
nonzero-sum
activity
that
there's
not
a
fixed
amount
of
investment,
my
share
of
development
contributions
and
that
working
together
we
actually
create
a
positive
some
scenario
that
works
much
better
than
any
of
us
working
on
our
own,
and
so
our
hope
is
that
going
forward
squeeze
sees
that
and
feels
that
and
then
finishing
up
slide.
24
I
would
really
encourage
you
to
reach
out
to
me
by
email
on
Twitter
or
via
slack.